Workshop on Public Transport - April 24, 2018

ECOPT intends to organize several small workshops with both a scientific speaker and a speaker from practice throughout the year. The next workshop takes place on Tuesday, April 24, 2018, at the Erasmus University Rotterdam.

Speakers and program

The speakers in this workshop are:

  • Valentina Cacchiani (Universit√† di Bologna)
  • Marcel van Kooten Niekerk (Utrecht University)

The program is as follows.

13:45 - 14:00 Welcome
14:00 - 15:00 Valentina Cacchiani Train Timetabling in Highly Congested Lines
15:00 - 15:30 Break
15:30 -16:30 Marcel van Kooten Niekerk Optimizing for Reliable and Sustainable Public Transport
From 16:30 Drinks


Valentina Cacchiani: Train Timetabling in Highly Congested Lines

Train Timetabling is a fundamental step to obtain an efficient use of railway networks. It deeply affects the passenger perception of the service provided and influences the Rolling Stock Circulation and Crew Scheduling. We present a time-space graph based Integer Linear Programming (ILP) model for solving the Train Timetabling problem. Most of the constraints are relaxed in a Lagrangian way, and a heuristic algorithm is proposed to solve the problem. We present an application of the proposed heuristic algorithm to the problem of scheduling additional trains in highly congested lines, in which Train Timetabling and Stop Planning are combined.

Marcel van Kooten Niekerk: Optimizing for Reliable and Sustainable Public Transport

In this presentation, we will discuss three different subjects.

In the first part, we present the improvement we made to current trip runtime determination methods. In our model, we distinguish between holding points(stops where vehicles wait until planned departure time) and non-holding points. Furthermore, we also take passengers into account by considering average passenger travel and waiting time along with the traditional average delay and punctuality. Solving our model by using an ILP makes the model intractable for medium to large size instances, so we investigated other optimization techniques to optimize the planned trip runtimes.

In the second part, we consider the problem of improving punctuality and robustness in vehicle scheduling. Traditionally, minimum trip layovers are applied to prevent propagation of disruptions; these minimum layovers typically have the same value for a lot of trips. To improve upon this, we first propose a method to calculate the punctuality of a vehicle schedule taking propagation of disruptions and historical runtime measurements into account. After this, we propose and evaluate several methods for vehicle scheduling using measured trip runtimes, where we minimize the average delay, taking into account the possible propagation of disruptions. We show that using our methods instead of using predefined minimum layovers gives a vehicle schedule with a better punctuality for the same cost.

In the third part, we consider Electric Vehicles (EVs). Traditional vehicle schedule optimization often leads to infeasible schedules for EVs since the batteries are not big enough to enable a whole day of operation without recharging; therefore we take the properties of the batteries of the EVs into account at the time of vehicle scheduling. At the moment, in Europe the number of EVs in many networks is relatively small, so this can still be done manually. However, with the increasing popularity of these EVs, this will not be true anywhere in the near future. We propose two models and three solution techniques for the optimization of vehicle schedules for EVs. Each model has a different trade-off between problem size and quality of solution.

Practical information

Venue: EUR campus Woudestein, Polak Building Y1-21. Directions on how to get to the campus and a map of the campus can be found here

Registration: Participation in the workshop is free. However, only a limited number of places are available. Therefore, we ask you to register for the workshop by sending an email to Registrations will be served on a first-come-first-serve basis.